Setting up Gentoo Linux on UTM on Apple Silicon

Posted on Nov 3, 2023

These are my notes on setting up a ‘headless’ Gentoo Linux installation on UTM on my Apple Silicon Mac, with the Apple Virtualization ‘backend’.

I followed the Gentoo handbook.

I’ve never used Gentoo before and I’ve come away pretty impressed. There’s something about having the cruft that other distros imposed stripped away, and doing all the setup yourself, that somehow makes things seem easier to me because I ended up understanding things a lot more. And the installed system ‘feels’ pared down and efficient.

There is not yet an official handbook for arm64 (“The arm and arm64 architectures are supported by the Gentoo project but do not yet have Handbooks at their disposal due to too many variations in SoCs.”), but I found that the steps - for UTM and Apple Virtualization at least - are basically exactly the same as the ones in the AMD64 (64-bit Intel) handbook.

This is a condensed transcript mostly for myself. If you choose to follow this, and need to know more, check out the handbook!

Without further ado, my notes:

Setting up Gentoo arm64:

Download arm64 “Minimal Installation CD” from https://www.gentoo.org/downloads/

UTM: File -> New, “Virtualize”, “Linux”, “Use Apple Virtualization”

Select the Gentoo “Minimal Installation CD” as the boot iso. Set memory and leave CPU cores at default. Specify a hard drive size.

Boot.

Set up root password, user, and turn on SSH - this is so that we can use the macOS terminal instead of the virtual console to control installation (so e.g. copy/paste works).

    livecd ~ # passwd
    
    livecd ~ # rc-service sshd start
    
    livecd ~ # ip address

Now, minimize the UTM window and SSH in in a regular Mac terminal:

    jamie@Jamies-MacBook-Air ~> ssh root@192.168.64.8

Work out what the hard drive is:

    livecd ~ # lsblk 
    NAME   MAJ:MIN RM   SIZE RO TYPE MOUNTPOINTS
    loop0    7:0    0 446.2M  1 loop /mnt/livecd
    sda      8:0    0 692.4M  1 disk /mnt/cdrom
    ├─sda1   8:1    0   104K  1 part 
    ├─sda2   8:2    0   2.8M  1 part 
    └─sda3   8:3    0 689.4M  1 part 
    vda    252:0    0    50G  0 disk 

Format the drive:

    livecd ~ # fdisk /dev/vda

Create a GPT:

    Command (m for help): g

    Created a new GPT disklabel (GUID: 1A47D934-85E8-594A-BF02-1F39C5CE6B2F).

Set up partitions. No swap (for now?), so just an EFI boot partition and a Linux partition:

    Command (m for help): p

    Disk /dev/vda: 50 GiB, 53687091200 bytes, 104857600 sectors
    Units: sectors of 1 * 512 = 512 bytes
    Sector size (logical/physical): 512 bytes / 512 bytes
    I/O size (minimum/optimal): 512 bytes / 512 bytes
    Disklabel type: gpt
    Disk identifier: 1A47D934-85E8-594A-BF02-1F39C5CE6B2F
    
    Command (m for help): n
    Partition number (1-128, default 1): 1
    First sector (2048-104857566, default 2048): 
    Last sector, +/-sectors or +/-size{K,M,G,T,P} (2048-104857566, default 104855551): +500M
    
    Created a new partition 1 of type 'Linux filesystem' and of size 500 MiB.
    
    Command (m for help): t
    Selected partition 1
    Partition type or alias (type L to list all): 1
    Changed type of partition 'Linux filesystem' to 'EFI System'.
    
    Command (m for help): n
    Partition number (2-128, default 2): 2
    First sector (1026048-104857566, default 1026048): 
    Last sector, +/-sectors or +/-size{K,M,G,T,P} (1026048-104857566, default 104855551): 104857566
    
    Created a new partition 2 of type 'Linux filesystem' and of size 49.5 GiB.
    
    Command (m for help): w
    The partition table has been altered.
    Calling ioctl() to re-read partition table.
    Syncing disks.

Set up the filesystems:

    livecd ~ # mkfs.vfat -F 32 /dev/vda1
    mkfs.fat 4.2 (2021-01-31)
    livecd ~ # mkfs.xfs /dev/vda2
    meta-data=/dev/vda2              isize=512    agcount=4, agsize=3244735 blks
             =                       sectsz=512   attr=2, projid32bit=1
             =                       crc=1        finobt=1, sparse=1, rmapbt=0
             =                       reflink=1    bigtime=1 inobtcount=1 nrext64=0
    data     =                       bsize=4096   blocks=12978939, imaxpct=25
             =                       sunit=0      swidth=0 blks
    naming   =version 2              bsize=4096   ascii-ci=0, ftype=1
    log      =internal log           bsize=4096   blocks=16384, version=2
             =                       sectsz=512   sunit=0 blks, lazy-count=1
    realtime =none                   extsz=4096   blocks=0, rtextents=0
    Discarding blocks...Done.

Mount the new partitions:

    livecd ~ # mount /dev/vda2 /mnt/gentoo
    livecd / # mkdir /gentoo/efi
    livecd / # mount /dev/vda1 /gentoo/efi

Check the date, handbook says it must be correct (don’t worry about timezone):

    livecd ~ # date
    Mon Oct 30 18:25:27 UTC 2023

Download the Stage 3 tarball for arm64 from https://www.gentoo.org/downloads/ using wget:

    livecd ~ # cd /mnt/gentoo
    livecd /mnt/gentoo # wget <TARBALL URL>

Untar the stage3 tarball:

    livecd /mnt/gentoo # tar xpvf stage3-*.tar.xz --xattrs-include='*.*' --numeric-owner

Set compile options:

    livecd /mnt/gentoo # vi /mnt/gentoo/etc/portage/make.conf

I set:

    COMMON_FLAGS="-march=native -Os -pipe"  
    ACCEPT_LICENSE="*"    

[No need to set MAKEOPTS - Portage now defaults to setting the number of threads to the value returned by nproc.]

Select mirrors:

    livecd ~ # mirrorselect -6 -i -o >> /mnt/gentoo/etc/portage/make.conf

Set up DNS for chroot (this is necessary because the DHCP was done in the non-chroot environment)

    livecd ~ # cp --dereference /etc/resolv.conf /mnt/gentoo/etc/

Chroot to new partition:

    livecd ~ # mount --types proc /proc /mnt/gentoo/proc
    livecd ~ # mount --rbind /sys /mnt/gentoo/sys
    livecd ~ # mount --make-rslave /mnt/gentoo/sys
    livecd ~ # mount --rbind /dev /mnt/gentoo/dev
    livecd ~ # mount --make-rslave /mnt/gentoo/dev
    livecd ~ # mount --bind /run /mnt/gentoo/run
    livecd ~ # mount --make-slave /mnt/gentoo/run
    livecd ~ # chroot /mnt/gentoo /bin/bash
    livecd / # source /etc/profile
    livecd / # export PS1="(chroot) ${PS1}" 

Sync to latest ebuild repo:

    (chroot) livecd / # emerge-webrsync

    ...
    
    (chroot) livecd / # emerge --sync

Read news (or, at least, mark it as read…):

    (chroot) livecd / # eselect news read

Select profile (or not, the default one was what I wanted):

    (chroot) livecd / # eselect profile list
    
    ...
    
    (chroot) livecd / # eselect profile <number>

Updating the @world set. Handbook says this is necessary.

    (chroot) livecd / # emerge --ask --verbose --update --deep --newuse @world

I can’t live without VIM…

    (chroot) livecd / # emerge --ask --verbose vim

Configure USE variable? I didn’t see a need to change anything.

Set up CPU flags:

    (chroot) livecd / # emerge --ask app-portage/cpuid2cpuflags
    (chroot) livecd / # cpuid2cpuflags
    (chroot) livecd / # echo "*/* $(cpuid2cpuflags)" > /etc/portage/package.use/00cpu-flags

Set up timezone (find zone in e.g. ls /usr/share/zoneinfo/)

    (chroot) livecd / # echo "America/Los_Angeles" > /etc/timezone
    (chroot) livecd / # emerge --config sys-libs/timezone-data

Configure locales:

    (chroot) livecd / # vim /etc/locale.gen

Uncomment:

        en_US ISO-8859-1
        en_US.UTF-8 UTF-8
    (chroot) livecd / # locale-gen

    (chroot) livecd / # env-update && source /etc/profile && export PS1="(chroot) ${PS1}"

Kernel configuration:

Get extra firmware (probably not necessary?)

    (chroot) livecd / # emerge --ask sys-kernel/linux-firmware

Set up kernel:

    (chroot) livecd / # emerge --ask sys-kernel/installkernel-gentoo
    
    (chroot) livecd / # emerge --ask sys-kernel/gentoo-kernel

Set up fstab:

    (chroot) livecd / # blkid

    (chroot) livecd / # vim /etc/fstab 

    UUID="41F9-7C9A"   /efi        vfat    defaults    0 2
    UUID="95d0078d-00f9-4b14-b03e-64a9742bf609"   /            xfs    defaults,noatime              0 1

Networking:

    (chroot) livecd / # echo "gentoo-arm64" > /etc/hostname
    (chroot) livecd / # emerge --ask net-misc/dhcpcd
    (chroot) livecd / # rc-update add dhcpcd default
    (chroot) livecd / # emerge --ask --noreplace net-misc/netifrc

Cron:

    (chroot) livecd / # emerge --ask sys-process/cronie
    (chroot) livecd / # rc-update add cronie default

Syslog (Note: syslog-ng seems to put logs in /var/log/messages)

    (chroot) livecd / # emerge --ask app-admin/syslog-ng
    (chroot) livecd / # rc-update add syslog-ng default
    (chroot) livecd / # emerge --ask app-admin/logrotate

SSH:

    (chroot) livecd / # rc-update add sshd default

Bash completion:

    (chroot) livecd / # emerge --ask app-shells/bash-completion

Timekeeping. Skipped this - won’t the virtual host provide good time?

    root #emerge --ask net-misc/chrony
    root #rc-update add chronyd default

FS tools:

    (chroot) livecd / # emerge --ask sys-fs/xfsprogs sys-fs/dosfstools
    
    This not available on arm64:
        (chroot) livecd / # emerge --ask sys-block/io-scheduler-udev-rules

Bootloader:

    (chroot) livecd / # echo 'GRUB_PLATFORMS="efi-64"' >> /etc/portage/make.conf
    (chroot) livecd / # emerge --ask --verbose sys-boot/grub
    (chroot) livecd / # grub-install --target=arm64-efi --efi-directory=/efi

    (chroot) livecd / # grub-mkconfig -o /boot/grub/grub.cfg

Set up a user:

    (chroot) livecd / # useradd -m -G users,wheel,audio -s /bin/bash jamie
    (chroot) livecd / # passwd jamie

SET THE ROOT PASSWORD

    (chroot) livecd / # passwd

OR, set up sudo:

    (chroot) livecd / # emerge --ask --verbose app-admin/sudo
    (chroot) livecd / # vi /etc/sudoers
    
        Uncomment line that starts %wheel

Done!

    (chroot) livecd / # exit
    livecd / # umount -l /mnt/gentoo/dev{/shm,/pts,}
    livecd / # umount -R /mnt/gentoo
    livecd / # shutdown -h now

Remove the live CD from the virtual machine and boot.

Remove old live-CD key for the IP address from

    jamie@Jamies-MacBook-Air ~> vim /Users/jamie/.ssh/known_hosts

SSH in to new virtual machine:

Set up Avahi (zeroconf/bonjour) - this will allow you to e.g. ssh gentoo-arm64.local from the Mac terminal. Note: this installs a load of stuff because nothing has yet used xml2man, which Avahi uses for its docs.

    jamie@gentoo-arm64 ~ $ sudo emerge --ask net-dns/avahi
    jamie@gentoo-arm64 ~ $ sudo rc-update add avahi-daemon default

Other notes:

The virtual machine seems always to boot form the virtual hard drive, even if the virtual CD is available. To reboot from the CD if you need to rescue something:

In GRUB command line

    set root=(hd0,gpt2)
    chainloader /efi/bootaa64.efi
    boot

Paths may be wrong, but Grub has an ’ls’ that you can look at things in:

    ls (hd0,gpt2)/efi

etc.

After boot, device names may be different - check with:

    livecd ~ # blkid

Setting up DistCC

One goal for me is to use the arm64 machine as a compile server for x86 machines.

We need the latest distcc, with a fix for https://github.com/distcc/distcc/pull/427, which is in ’testing’ and not yet released to stable gentoo.

    jamie@gentoo-arm64 ~ $ echo "sys-devel/distcc ~arm64" > /etc/portage/package.accept_keywords/distcc

Build with crossdev support:

    jamie@gentoo-arm64 ~ $ echo "sys-devel/distcc crossdev" > "/etc/portage/package.use/sys-devel:distcc"
    
    
    jamie@gentoo-arm64 ~ $ sudo emerge --ask distcc
    jamie@gentoo-arm64 ~ $ sudo emerge --ask crossdev

As root:

    gentoo-arm64 /home/jamie # mkdir -p /var/db/repos/crossdev/{profiles,metadata}
    gentoo-arm64 /home/jamie # echo 'crossdev' > /var/db/repos/crossdev/profiles/repo_name
    gentoo-arm64 /home/jamie # echo 'masters = gentoo' > /var/db/repos/crossdev/metadata/layout.conf
    gentoo-arm64 /home/jamie # chown -R portage:portage /var/db/repos/crossdev
    
    gentoo-arm64 /home/jamie # vim /var/db/repos/crossdev/metadata/layout.conf
    
        add "thin-manifests=true"
    

    gentoo-arm64 /home/jamie # mkdir /etc/portage/repos.conf
    gentoo-arm64 /home/jamie # vim /etc/portage/repos.conf/crossdev.conf
        
        [crossdev]
        location = /var/db/repos/crossdev
        priority = 10
        masters = gentoo
        auto-sync = no

    jamie@gentoo-arm64 ~ $ sudo crossdev --stable -t i486-pc-linux-gnu

    jamie@gentoo-arm64 ~ $ sudo vim /etc/conf.d/distccd
    
    DISTCCD_OPTS="${DISTCCD_OPTS} --log-level info --log-file /var/log/distccd.log"
    DISTCCD_OPTS="${DISTCCD_OPTS} --allow 192.168.0.0/16"

    jamie@gentoo-arm64 ~ $ sudo touch /var/log/distccd.log
    jamie@gentoo-arm64 ~ $ sudo chown distcc:root /var/log/distccd.log
    
    jamie@gentoo-arm64 ~ $ sudo rc-update add distccd default

That’s is - despite many web pages detailing insane complexity of using distcc with crossdev, nowadays it ‘just works’ after installation.